THI22

Summary: Phylogenetically inferred cytosolic localization, consistent with the thiamine-pathway biochemistry of the paralogs THI20/THI21, which act in the cytosolic HMP-P kinase steps. Plausible as a location but not a core function and not experimentally shown for THI22.

Reason: IBA location annotation from the THI20-related PANTHER family. The de novo thiamine biosynthesis kinase steps occur in the cytosol, so cytosolic localization is the reasonable default for a family member. It conflicts with the InterPro/SubCell "extracellular region" prediction (see below); the cytosolic assignment is the more biologically defensible of the two but remains inferred, so it is retained as non-core.

Summary: HMP kinase activity (HMP -> HMP-P, EC 2.7.1.49) is a demonstrated activity of the family (THI20 is trifunctional). THI22 retains the N-terminal kinase domain and its HMP substrate-binding residue (Gln, conserved with THI20 Q64), so the activity is plausible on sequence grounds, but it is inferred by phylogeny and not demonstrated for THI22.

Reason: Domain-defensible family activity (N-terminal ThiD-like kinase domain; substrate-binding and Gly-rich phosphate-binding motifs conserved in THI22 per inline alignment to THI20). Kept as non-core because THI22-specific activity is unproven; the one direct enzymatic assay of the family that tested THI22 failed to detect the related HMP-P kinase activity (per UniProt CAUTION, ECO:0000305, and PMID:10383756, which demonstrated the activity only for THI20/THI21).

Summary: HMP-P kinase activity (HMP-P -> HMP-PP, EC 2.7.4.7) is exactly the activity that PMID:10383756 demonstrated for two of the three family members and could NOT demonstrate for the third (THI22). Retained by phylogenetic inference but flagged as over-annotated for THI22 because it is directly contradicted by the one experiment that tested it.

Reason: The IBA propagates the well-established HMP-P kinase activity of THI20/THI21 onto THI22, but the primary paper demonstrated this activity for only two of the three members and UniProt records (CAUTION, ECO:0000305) that no HMP-P kinase activity could be demonstrated for THI22 despite its high homology. The catalytic residues are conserved (so this is not a residue-dead pseudoenzyme and REMOVE is not justified), but calling this a THI22 function over-states the evidence: it is a family-level inference contradicted by the only direct THI22 assay.

Summary: Thiamine biosynthetic process is the family process, but THI22 has been shown NOT to be required for thiamine biosynthesis. The phylogenetic propagation over-states THI22's role in de novo biosynthesis.

Reason: UniProt (from PMID:10383756) states THI22 "is not required for thiamine biosynthesis". The IBA correctly captures the family process but over-annotates THI22, whose contribution to de novo biosynthesis is unproven and, on deletion evidence in the redundant family, dispensable. A broader thiamine metabolic-process framing is more defensible for THI22 (see GO:0006772).

Summary: This location is derived automatically from a predicted N-terminal signal peptide (UniProt SubCell "Secreted", ECO:0000305). The signal is probably genuine, but the destination is wrong - the supported compartments are endomembrane (ER, vacuole), not the extracellular space.

Reason: Pure prediction from a signal-peptide/SubCell mapping, not experimental; no evidence places THI22 outside the cell, and SGD did not accept an extracellular annotation (GO:0005575 ND). The verdict stands, but the earlier rationale for it was wrong and has been rewritten after a blinded OpenScientist run (2026-08-08; see the file: reference). That run independently reproduced the held-out local observation that the N-terminal extension is unique to THI22 among THI20/THI21/THI22, quantified it (hydropathy peak 3.78 vs ~0.5-0.8 in the paralogs), and then read it the opposite way to this review: the extension is most likely a functional ER-targeting signal, and the paralog contrast is evidence of neofunctionalization toward the endomembrane system rather than evidence that THI22 is cytosolic like its sisters. The error in GO:0005576 is therefore the endpoint, not the signal - entering the secretory pathway is not the same as being secreted. SGD's locus record states that SWAT-GFP and mCherry fusions localize to the endoplasmic reticulum and vacuole respectively. Those compartments are recorded as leads only, not annotated here, because the primary imaging references are not cached and the calls are high-throughput and tag-dependent (N-terminal vs C-terminal tags disagree). This also puts the cytosol IBA on this gene in question; it is left untouched pending the same primary references.

Summary: General thiamine metabolic process (parent of both biosynthesis and salvage/degradation). Defensible at the family level: THI22 is thiamine-regulated and carries intact HMP-P kinase and thiaminase-II-like domains. This broad process term over-commits less than the specific biosynthesis term.

Reason: InterPro-based BP annotation. Appropriate as a broad, non-core framing given THI22's thiamine-responsive regulation and its two thiamine-metabolism domains, while avoiding the stronger and contradicted "biosynthetic process" claim.

Summary: InterPro-derived duplicate of the HMP-P kinase MF annotation. Same assessment as the IBA HMP-P kinase annotation: family activity propagated to THI22 but directly not demonstrated for THI22.

Reason: InterPro (IPR004399) maps the ThiD-like kinase domain to HMP-P kinase activity. The domain is present and its catalytic residues are conserved, but the specific HMP-P kinase activity was tested for THI22 and not detected (UniProt CAUTION, ECO:0000305; PMID:10383756 demonstrated the activity only for THI20/THI21), so this electronic annotation over-annotates THI22.

Summary: Automated (multi-method IEA) thiamine biosynthetic process annotation. Same over-annotation concern as the IBA biosynthesis annotation: THI22 is not required for thiamine biosynthesis.

Reason: ARBA/InterPro electronic inference of de novo biosynthesis involvement, contradicted for THI22 by the functional characterization showing it is not required for thiamine biosynthesis. Broader thiamine metabolic process (GO:0006772) is the defensible level.

Summary: Thiaminase II activity (thiamine hydrolysis to HMP + thiazole, EC 3.5.99.2) is a demonstrated activity of THI20 and maps to the C-terminal TenA/thiaminase-II-like domain (IPR027574). THI22 retains this domain and both catalytic residues (nucleophile Cys and proton-donor Glu, conserved with THI20 C468/E540), so the activity is domain-plausible but not demonstrated for THI22.

Reason: Domain-defensible family activity (C-terminal TenA/thiaminase-II domain with conserved catalytic Cys/Glu per inline alignment). Kept as non-core rather than accepted as a core function because no thiaminase activity has been directly demonstrated for THI22, and the 1999 paper noted the C-terminal domain's function was unresolved.

Summary: Root molecular_function placeholder (ND, no biological data).

Reason: ND root-term placeholder recording that no specific molecular function had been curated by SGD at annotation time. Standard practice; nothing to change.

Summary: Root cellular_component placeholder (ND, no biological data).

Reason: ND root-term placeholder recording that no specific localization had been curated by SGD at annotation time. Standard practice; nothing to change.

Summary: SGD experimental (IEP) annotation from the primary paper. The underlying evidence is that THI22 expression is induced under thiamine limitation, coordinately with THI20/THI21. This thiamine-responsive regulation is genuinely established; however, the same work shows THI22 is not required for de novo thiamine biosynthesis, so annotation of THI22 to the specific "biosynthetic process" term over-states its role — the co-regulation evidence supports thiamine-pathway involvement, not a demonstrated biosynthetic contribution.

Reason: IEP (inferred from expression pattern) reflects real, curator-read experimental evidence of thiamine-dependent regulation, and per curation guidelines the experimental evidence itself is not discarded. But the specific target term (thiamine BIOSYNTHETIC process) over-states THI22's role: the same paper establishes THI22 is not required for thiamine biosynthesis. The regulation-based involvement is better captured by the broader thiamine metabolic process (GO:0006772). Marked over-annotated (not removed) to preserve the experimental co-regulation evidence while flagging the term as too specific for THI22.